Suppressors of Superoxide Dismutase (SOD1) Deficiency in Saccharomyces cerevisiae

Yeast deficient in the cytosolic copper/zinc superoxide dismutase (SOD1) exhibit metabolic defects indicative of oxidative damage even under non-stress conditions. To help identify the endogenous sources of this oxidative damage, we isolated mutant strains of S. cerevisiae that suppressed metabolic defects associated with loss of SOD1. Six complementation groups were isolated and three of the corresponding genes have been identified. One sod1Δ suppressor represents SSQ1 which encodes a hsp70-type molecular chaperone found in the mitochondria. A second sod1Δsuppressor gene, designated JAC1, represents a new member of the 20-kDa J-protein family of co-chaperones. Jac1p contains a mitochondrial targeting consensus sequence and may serve as the partner for Ssq1p. Homologues of Ssq1p and Jac1p are found in bacteria in close association with genes proposed to be involved in iron-sulfur protein biosynthesis. The third suppressor gene identified wasNFS1. Nfs1p is homologous to cysteine desulfurase enzymes that function in iron-sulfur cluster assembly and is also predicted to be mitochondrial. Each of the suppressor mutants identified exhibited diminished rates of respiratory oxygen consumption and was found to have reduced mitochondrial aconitase and succinate dehydrogenase activities. Taken together these results suggest a role for Ssq1p, Jac1p, and Nfs1p in assembly/maturation of mitochondrial iron-sulfur proteins and that one or more of the target Fe/S proteins contribute to oxidative damage in cells lacking copper/zinc SOD.

Yeast deficient in the cytosolic copper/zinc superoxide dismutase (SOD1) exhibit metabolic defects indicative of oxidative damage even under non-stress conditions. To help identify the endogenous sources of this oxidative damage, we isolated mutant strains of S. cerevisiae that suppressed metabolic defects associated with loss of SOD1. Six complementation groups were isolated and three of the corresponding genes have been identified. One sod1⌬ suppressor represents SSQ1 which encodes a hsp70-type molecular chaperone found in the mitochondria. A second sod1⌬ suppressor gene, designated JAC1, represents a new member of the 20-kDa J-protein family of co-chaperones. Jac1p contains a mitochondrial targeting consensus sequence and may serve as the partner for Ssq1p. Homologues of Ssq1p and Jac1p are found in bacteria in close association with genes proposed to be involved in iron-sulfur protein biosynthesis. The third suppressor gene identified was NFS1. Nfs1p is homologous to cysteine desulfurase enzymes that function in iron-sulfur cluster assembly and is also predicted to be mitochondrial. Each of the suppressor mutants identified exhibited diminished rates of respiratory oxygen consumption and was found to have reduced mitochondrial aconitase and succinate dehydrogenase activities. Taken together these results suggest a role for Ssq1p, Jac1p, and Nfs1p in assembly/ maturation of mitochondrial iron-sulfur proteins and that one or more of the target Fe/S proteins contribute to oxidative damage in cells lacking copper/zinc SOD.
Superoxide dismutase (SOD) 1 (3)(4)(5)(6)(7). This oxidation of [4Fe-4S] clusters liberates iron that is then free to catalyze Fenton chemistry production of highly toxic hydroxyl radicals (OH) (8,9). The determinants of oxidative damage relevant to eukaryotic SOD are less well understood. Eukaryotes typically contain a manganese-containing SOD (SOD2) that is confined to the matrix of the mitochondria and a copper/zinc-containing SOD1 that appears to be widely distributed throughout the cell, but is excluded from the mitochondrial matrix (10 -12). Genetic studies in the bakers' yeast Saccharomyces cerevisiae have shown that mitochondrial SOD2 acts in defense against respirationderived radicals. Yeast cells lacking SOD2 ("sod2⌬") are sensitive toward environmental oxidants (e.g. paraquat) and toward hyperoxia (13); in these sod2⌬ mutants oxygen sensitivity can be suppressed by rho o mutations in mitochondrial DNA that block respiration (14). The role of copper/zinc SOD1 appears to be more complex. Yeast strains lacking SOD1 are also sensitive toward environmental oxidants and hyperoxia, but unlike sod2⌬ cells, the oxygen sensitivity of sod1⌬ yeast is not corrected by rho o mutations blocking respiration (15). In addition, sod1⌬ mutants exhibit markers of oxidative damage that appear to arise from endogenously generated oxidants. These sod1⌬ strains exhibit air-dependent blockages in methionine and lysine biosynthesis that are not observed with sod2⌬ strains (16 -20). Moreover, both auxotrophies are observed under normoxic conditions in the absence of additional insult from environmental oxidants (16 -20), suggesting that these metabolic defects in sod1⌬ cells arise as a product of metabolismderived oxygen radicals. Valentine and colleagues (21) have shown that inactivation of coenzyme Q partially alleviates the amino acid auxotrophies of sod1⌬ cells, suggesting that SOD1 does play a role in protection from O 2 . produced by mitochondrial respiration.
In an attempt to better understand the determinants of oxygen toxicity related to SOD1, we sought to identify gene mutations that specifically suppress the amino acid auxotrophies of sod1⌬ yeast as markers of metabolic oxidative damage. This study led to the isolation of three nuclear genes that are predicted to participate in the assembly of mitochondrial ironsulfur clusters. Our findings are consistent with a model in which SOD1 helps guard against damage derived from mitochondrial iron-sulfur proteins.
Stocks of yeast strains were maintained on YPD medium containing 2% glucose and tests for methionine and lysine auxotrophy were carried out using a synthetic dextrose medium (27). Succinate dehydrogenase assays were conducted on cells grown in a modified YPD medium containing 0.6% glucose. Anaerobic cultures were maintained by growth in a O 2 depleted culture jar (BBL Gas Pak). Genetic analysis was conducted according to standard procedures (27).
Isolation and Characterization of seo Mutants-The parental strain used for isolation of suppressors of SOD1 deficiency was the sod1⌬::TRP1 strain KS105 transformed with the PMR1 LEU2 CEN plasmid p74C (28). An extra copy of the PMR1 gene was introduced in this strain to preclude the isolation of pmr1 mutational suppressors that typically accumulate to a very high frequency (10 Ϫ5 -10 Ϫ6 ) in yeast lacking SOD1 (16). Ten independent PMR1 transformants of KS105 were plated onto medium lacking either lysine or methionine, and following 4 days of aerobic growth, methionine, or lysine prototrophic colonies appeared at a frequency of 10 Ϫ7 . A large number of these were isolated and Leu Ϫ derivatives were generated that lost the PMR1 LEU2 plasmid. Clones that consistently exhibited methionine or lysine prototrophy and lack of growth on medium containing 0.5 mM paraquat were chosen for further study. These were designated as seo (suppressors of endogenous oxygen toxicity) mutants. Twenty independent seo clones were mated to the sod1⌬ strain of opposite mating type (JS005), and 10 recessive mutants were identified. The diploids were sporulated and in all cases, methionine/lysine prototrophy exhibited 2:2 segregation, demonstrating mutation of a single gene. Utilizing MAT␣ seo segregants, complementation analysis was conducted defining six seo complementation groups. seo1 is represented by three alleles, seo5 and seo6 each have 2 members and the remainder seo mutants are represented by single alleles.
Gene Cloning and Molecular Biology Techniques-To isolate the wild type chromosomal genes corresponding to seo1, seo2, and seo4, the appropriate seo strain was transformed with a wild type genomic DNA library present on a LEU2 CEN vector (kind gift of F. Spencer, Johns Hopkins University). Approximately 15,000 Leu ϩ colonies were replicaplated onto two sets of leucine Ϫ plates that either contained or lacked methionine. Following 2 days of aerobic growth, methionine auxotrophic colonies were identified. Plasmids were rescued from these candidates and the genomic DNA insert of those plasmids that consistently showed complementation of the seo1, seo2, or seo4 methionine prototrophy were subjected to partial DNA sequence analysis. A search of the S. cerevisiae data base revealed the open reading frames contained within each. Subcloning and deletion analysis demonstrated that the complete open frames of SSQ1, YGL018C (referred to herein as JAC1), and NFS1 were necessary and sufficient to complement the seo1, seo2, and seo4 mutants, respectively.
To construct the yfh1⌬::URA3 plasmid pJS406, YFH1 sequences Ϫ480 to ϩ15 and ϩ160 to ϩ671 were amplified by PCR using modified primers that introduced a SalI site at ϩ160 or a BamHI site at ϩ15. The products of PCR were digested with HindIII and SalI (downstream fragment) or HindIII and BamHI (upstream fragment) and ligated in a trimolecular reaction to the SalI and BamHI sites of the URA3 integrating vector, pRS306. The resultant pJS406 plasmid was linearized with HindIII and used to delete the chromosomal YFH1 gene of KS105.
Measurements of Oxygen Consumption-Yeast strains grown to early stationary phase in YPD medium were harvested and resuspended to a density of 1 ϫ 10 9 cells/ml in a phosphate-buffered solution containing 2% glucose. A 0.5-ml aliquot was added to 2.5 ml of phosphate-buffered saline for which an oxygen-saturation baseline had been established. Oxygen consumption per unit time was then recorded continuously through use of a Clark oxygen electrode. Following several minutes of monitoring, KCN was added at a final concentration of 1 mM and recording continued for an additional 4 -5 min. The KCN inhibitable respiration rate was then determined as a function of differential slopes obtained.
Enzyme Assays-Assays for aconitase were conducted essentially as described (29). Cells were grown in YPD medium to stationary phase, harvested by centrifugation, and broken by glass bead homogenization in phosphate buffer (pH 7.4). Cell lysates containing 75 g of protein were assayed at 25°C in 500 l of aconitase reaction buffer (29) containing 20 mM Tris-HCl (pH 7.4), 100 mM NaCl, and 0.5 mM cisaconitate. The decrease in absorbance at 240 nm was measured as a function of time and an extinction coefficient of 4.88 (mM cm) Ϫ1 was utilized to calculate activity; 1 unit of aconitase activity is defined as 1 nmol of cis-aconitate converted/min/mg of protein.
Assays for succinate dehydrogenase were carried out as described (30). Cells were grown in YPD (0.6% glucose) medium to stationary phase, harvested by centrifugation, suspended in 20 mM Hepes (pH 7.4) containing 1 mM phenylmethylsulfonyl fluoride, and broken by passage through a French pressure cell three times at 20,000 psi. Cell lysates were clarified by centrifugation at 10,000 ϫ g for 10 min, and submitochondrial particles were isolated by centrifugation at 100,000 ϫ g for 60 min. Activities were determined at 30°C in 50 mM Hepes (pH 7.4), 0.1 mM EDTA, 1 mM potassium cyanide, 100 M phenazine methosulfate, and 20 mM succinate by reduction of dichlorophenol indophenol, ⌬⑀ ϭ Ϫ21 (mM cm) Ϫ1 . Activities reported reflect linear initial rates and are directly proportional to amounts of added protein.

Isolation of seo Mutants (Suppressors of Endogenous Oxygen
Toxicity)-In an attempt to identify metabolic sources of oxidative damage, S. cerevisiae gene mutations were sought that suppressed oxygen toxicity in a yeast strain lacking copper/zinc SOD1. Two classes of gene mutations were expected from this analysis: mutations that resulted in an increased "anti-oxidant" capacity of the cell and those that dampened the "prooxidant" production of oxygen radicals from metabolic sources. Mutations that act at the pro-oxidant level were expected to suppress only markers of endogenous oxygen toxicity (e.g. the methionine or lysine auxotrophy) and not the sensitivity of sod1⌬ yeast toward environmental oxidants (e.g. paraquat). Therefore, we specifically isolated mutants that permitted aerobic growth of a sod1⌬ strain on medium lacking methionine or lysine but did not alleviate the sensitivity of this strain toward paraquat. These yeast isolates were designated as seo mutants (suppressors of endogenous oxygen toxicity). Genetic analysis defined those seo gene mutations that were recessive, and these were classified into six complementation groups (see "Experimental Procedures"). As shown in Fig. 1, all six seo mutants rescued to some degree both the lysine and methionine auxotrophy of the sod1⌬ parental strain.
Because the mitochondrial respiratory chain represents a potential source of oxidative damage, we investigated whether the seo mutants were respiratory deficient. The inability to utilize non-fermentable carbon sources is a hallmark of respiratory deficiency in yeast. When plated onto medium containing ethanol and glycerol, mutants seo1 and seo4 exhibited growth, albeit at somewhat reduced rates, whereas mutants seo2, seo3, seo5, and seo6 exhibited negligible growth under these conditions (not shown). To more precisely monitor respiration, measurements of cyanide-inhibitable oxygen consumption were obtained. As seen in Fig. 1, all of the seo groups exhibited diminished respiration. The seo2 and seo5 mutants showed negligible rates (Ͻ10%), seo3 and seo6 showed rates Ͻ20%, and seo1 and seo4 exhibited oxygen consumption rates ranging from 50 to 80% that of the corresponding wild type parental strain. These studies suggest that alterations in respiration linked electron transfer may play a role in the suppression of oxidative damage in certain classes of seo mutants.
seo1 Is Equivalent to SSQ1, a Mitochondrial 70-kDa Molecular Chaperone-A wild type genomic DNA library was screened for clones that complemented the aerobic methionine prototrophy of the sod1⌬ seo1 strains. These studies revealed that the S. cerevisiae SSQ1 gene 2 was necessary and sufficient for complementation of the seo1 mutants. The Ssq1 protein was previously identified as a member of the hsp70 molecular chaperone family that localizes to mitochondria (31). SSQ1 is not an essential gene, but the null mutant exhibits a cold-sensitive phenotype (31). To confirm that the seo1 complementation group represented mutations in SSQ1, the complete open reading frame of SSQ1 was sequenced in three independent alleles of seo1 that were isolated. All three seo1 isolates contained nonsense mutations that delete portions of the C-terminal region of Ssq1p. As shown in Fig. 2, ssq1-1 contains a single nonsense mutation at Lys-587, ssq1-2 contains two mutations, a Ser to Pro substitution at position 594 and a nonsense mutation at Gln-611, and ssq1-3 harbors a single nonsense mutation at Gln-611. SSQ1 from the parental sod1⌬ strain was found to encode the full-length protein of 657 amino acids. Alignment of the sequence of Ssq1p to the crystal structure of the bacterial hsp70, DnaK (32), reveals that these mutations occur within a ␣-helical subdomain of the peptide-binding region. This subdomain does not make contacts with bound peptide, but it has been shown to affect polypeptide binding affinity (33) and to mediate interactions with regulatory proteins (34) in mammalian hsc70. Yeast strains containing a ssq1⌬ null mutation exhibit negligible growth at 30°C (31), whereas each of the seo1 mutants are able to grow at this temperature. This suggests that truncation of 45-70 residues from the C terminus of Ssq1p does not fully inactivate the protein, but rather "cripples" it sufficiently to suppress oxidative damage in sod1⌬ mutants.
Very recently, SSQ1 was identified in a screen for yeast genes that constitutively activate iron uptake pathways (35). These authors noted that strains defective for Ssq1p exhibit abnormal processing of mitochondrial Yfh1p, the yeast frataxin homologue involved in mitochondrial iron homeostasis (36 -39). To address whether ssq1 mutations suppress oxidative damage by impairing Yfh1p, we constructed a yfh1⌬ sod1⌬ strain. As seen in Fig. 3, loss of Yfh1p function does not mimic a ssq1 mutation, as a yfh1⌬ allele fails to suppress the aerobic methionine auxotrophy of sod1⌬ strains. Therefore the suppression of sod1⌬ by ssq1 mutations cannot be simply explained by a defect in Yfh1p.
seo2 Is Equivalent to a New Member of the J-family of Molecular Chaperones-The gene corresponding to the seo2 mutant was cloned in a manner analogous to that described for SSQ1. These studies revealed that complementation of seo2 required a single open reading frame (YGL018C) predicted to encode a polypeptide of 21.8 kDa. This protein appears homologous to a subfamily of J-domain containing co-chaperone proteins ϳ20 kDa, and we have designated this gene JAC1 (J-type accessory chaperone; see "Discussion"). Fig. 4A shows a comparison of the amino acid sequence of the Jac1 protein with the E. coli homologue, Hsc20. Jac1p has an N-terminal extension not present in the bacterial protein, and this region has features consistent with a role as a mitochondrial targeting sequence (40). Residues 26 -82 share similarities with J-motif sequences including the His-Pro-Asp signature sequence at positions 48 -50 (see Ref. 41). The C-terminal portion of Jac1p and E. coli Hsc20 is dissimilar from that of the larger DnaJ/ hsp40 proteins (ϳ40 kDa) and appears to comprise a domain unique to this class of co-chaperone proteins.
Sequence analysis of the seo2 suppressor revealed a triplet deletion mutation in JAC1 resulting in loss of the GAT codon for Asp-32 without disrupting the reading frame (Fig. 4B). Asp-32 is highly conserved in this subfamily of co-chaperones. To determine whether the triplet deletion represents a null mutation, we examined the effects of deleting the complete open reading frame of JAC1. An ura3⌬0 diploid harboring one wild type and one jac1⌬::URA3 allele was induced to sporulate, and tetrads were assayed for growth at 23, 30, and 37°C. Spore viability segregated 2:2 at all three temperatures (Fig. 4C), and all viable spores were auxotrophic for uracil (not shown). Thus, JAC1 is essential for S. cerevisiae growth, and the triplet deletion mutation is presumed to suppress sod1⌬ through partial disruption of Jac1p function.
seo4 Is Equivalent to NFS1-Gene cloning experiments revealed that seo4 is identical to S. cerevisiae NFS1, a gene encoding a protein homologous to bacterial NifS proteins (42). The NifS proteins, originally discovered in nitrogen fixing organisms, exhibit cysteine desulfurase activity and have been implicated in providing sulfide for assembly of iron-sulfur clusters of nitrogenase components (43,44). NifS-like proteins have since been found to occur widely in non-nitrogen fixing organisms, and demonstrations that Azotobacter vinelandii NifS can support assembly of heterologous Fe/S proteins (45)(46)(47) suggest that the NFS1 gene product and its homologues function generally in Fe/S cluster assembly. An alignment of yeast Nfs1p with the E. coli homologue is shown in Fig. 5. Like Ssq1p and Jac1p, the Nfs1 protein contains an N-terminal extension with features consistent with a role as a mitochondrial targeting sequence (40) that is absent in the bacterial homologue. Apart from this extended leader sequence, the yeast and bacterial proteins share ϳ60% amino acid sequence identity.
To confirm that mutation of the NFS1 gene was responsible for suppressing SOD1 deficiency, the complete NFS1 open reading frames of seo4 and the wild type parent were sequenced. As shown in Fig. 5, a single Ala-308 to Thr replacement was noted in the seo4 mutant, but not in the wild type protein. The exact role of Ala-308 is not known, but it occurs within a region of amino acids that is highly conserved among NifS homologues from both nitrogen fixing and non-fixing organisms and is thus likely to be critical for enzyme function. Because the nfs1 null mutant is enviable (48,49), this nfs1 substitution mutation is expected only to partially inactivate FIG. 1. Suppression of SOD1 deficiency in six seo complementation groups. Yeast strains were tested for aerobic methionine and lysine auxotrophy by plating 5 ϫ 10 5 cells in a 10-l solution onto synthetic dextrose medium supplemented with methionine or lysine as indicated. Aerobic growth at 30°C proceeded for 3 days. sod1⌬, parent strain KS105; remaining strains are seo derivatives of KS105, as indicated. resp. rate, rate of cyanide inhibitable oxygen consumption shown as a percentage of that obtained with the parental sod1⌬ strain. n, negligible rates of oxygen consumption (Ͻ10%). Each value represents the average of four to five experimental trials, conducted in triplicate; values in parentheses show ranges observed (Ϯ%). The standard deviation of the result obtained with the sod1⌬ seo1 mutant was 5.7%. protein function as suggested for the other seo mutations, jac1 and ssq1.
Activities of Mitochondrial Fe/S Enzymes-Because Nfs1p is predicted to function in the maturation of mitochondrial ironsulfur proteins, we tested whether two mitochondrial Fe/S enzymes, aconitase and succinate dehydrogenase, were affected in nfs1 mutants and in the other seo suppressor strains as well. Because the labile [4Fe-4S] cluster of aconitase is prone to oxidation by O 2 . (3,50), we first tested whether the sod1⌬ mutation present in the seo strains had any effect on aconitase activity. As shown in Fig. 6, sod1⌬ strains had roughly 50% of the aconitase activity of the corresponding wild type strain; a similar effect was produced by a sod2⌬ mutation which deletes the mitochondrial manganese Sod2p. These results are consistent with what has been reported for SOD-deficient E. coli cells (3). To eliminate effects of SOD deficiency, we re-introduced the yeast SOD1 gene into the seo mutants by transformation with a SOD1 CEN plasmid. Fig. 6 shows that the sod1⌬ strain harboring this episomal copy of SOD1 exhibits wild type levels of aconitase activity. In contrast, each of the seo mutants carrying the SOD1 CEN plasmid to compensate for the sod1⌬ mutation exhibits greatly diminished rates of aconitase activity suggestive of a possible defect in Fe/S cluster assembly. Similar studies were carried out to determine whether the activity of succinate dehydrogenase was affected in the suppressor mutants (Fig. 7). Each of the suppressor strains, with the exception of seo4, exhibits reduced activity. Thus, mutations in SSQ1 (seo1) and JAC1 (seo2) greatly reduce succinate dehydrogenase activity, but the NFS1 mutation has no effect. The lack of an effect of nfs1-1 on succinate dehydrogenase compared with the large effect observed for aconitase may reflect differences in the stability of the Fe/S clusters of the two enzymes (3,5,50,51). The impaired rate of sulfide production by the crippled, but not inactive, Nfs1-1 protein may be to sufficient to support maturation and repair of succinate dehydrogenase but insufficient to maintain aconitase in an active form.

DISCUSSION
The goal of these studies was to identify factors contributing to an oxidative stress state in yeast cells lacking copper/zinc SOD1. Six complementation groups were identified that suppress metabolic oxidative damage in sod1⌬ mutants. In earlier studies, we identified bsd suppressors of sod1⌬ that act at the antioxidant level to combat toxicity from both endogenous and environmental sources of free radicals (28, 52). In contrast, the seo mutations described here show no general antioxidant behavior in that they fail to ameliorate toxicity from environmental oxidants such as paraquat. Rather, the seo mutants appear to act at the pro-oxidant level to dampen one or more endogenous source of oxidative damage.
The genes affected in three of the seo complementation groups have been cloned. Two of these genes (SSQ1 and JAC1) encode members of the molecular chaperone family, and the third (NFS1) is a member of the NifS family of proteins involved in iron-sulfur cluster assembly. Several intriguing parallels can be drawn with these three genes. First, all three appear to encode mitochondrial proteins. Ssq1p has been shown to be mitochondrial (31), and our preliminary cell fractionation studies indicate a mitochondrial localization for Jac1p (not shown). The presence of a mitochondrial targeting consensus sequence for Nfs1p suggests a mitochondrial localization for this protein as well. A second feature common to JAC1, SSQ1, and NFS1 is that each are required for normal yeast growth. The jac1 and nfs1 null mutants are not viable. The ssq1 null mutant is viable but grows only at elevated temperature where increased expression of another mitochondrial chaperone, Ssc1p, may compensate for loss of function (31). The mutant alleles isolated in this study can thus be expected only to partially inactivate the encoded proteins, this reduction being sufficient to suppress oxidative damage in yeast lacking SOD1. As a third common denominator for JAC1, SSQ1, and NFS1, all three appear necessary for functioning of mitochondrial Fe/S proteins. Mutations in these genes result in impairment of respiration and reduction in activities of the mitochondrial Fe/S-containing enzymes, aconitase and succinate dehydrogenase.  3. Effects of a yfh1 versus ssq1 mutations on the aerobic  methionine auxotrophy of a sod1⌬ strain. The indicated yeast strains were tested for methionine auxotrophy by plating 5 ϫ 10 5 or 1 ϫ 10 5 cells in a 10-l solution onto synthetic dextrose medium supplemented with lysine but lacking methionine. Growth at 30°C proceeded for 3 days either in air (ϩoxygen) or in an oxygen-depleted culture jar (Ϫoxygen). Strains utilized are all derived from KS105 (sod1⌬) and harbor either the designated mutant alleles of SSQ1 or a null mutation in YFH1, as indicated.
FIG. 5. The S. cerevisiae Nfs1 protein and nfs1-1 mutant. The predicted amino acid sequence of the S. cerevisiae Nfs1p is aligned with the homologous protein (IscS) from E. coli. The Ala-308 to Thr mutation identified in the single allele of the seo4 complementation group is identified by an arrow. A potential leader sequence for mitochondrial targeting is denoted as described in legend to Fig. 4A.   FIG. 4. The S. cerevisiae Jac1 protein and jac1-1 mutant. A, the predicted amino acid sequence of S. cerevisiae Jac1p is aligned with the E. coli hscB gene product, Hsc20. Bold faced numbers designate amino acid position. Amino acid identities are indicated by asterisks; similarities are marked by a dot. The arrow points to Asp-32, deleted in the jac1-1 mutant. A potential mitochondrial leader sequence is marked by छ (hydrophobic), ϩ (acidic), and E (hydroxylated) residues. B, comparison of wild type (parental strain KS107) and mutated (seo2-1 derivative) JAC1 alleles surrounding the triplet deletion mutation (blocked). C, tetrad analysis of a JAC1/jac1⌬ heterozygous diploid: each panel shows the growth patterns of four representative tetrads resulting from sporulation of a JAC1/jac1⌬ heterozygous diploid. Circles on the left indicate placement of each of the four spores following dissection of a tetrad. Spore growth proceeded for 3 days at the indicated temperature. Viable spores from a large number of tetrads were plated onto medium lacking uracil and all were found to be auxotrophic for uracil. chaperone genes and genes homologous to the Fe/S cluster maturation genes (nif genes) of nitrogen fixing bacteria. An analogous gene cluster, separate from the nif genes, was also recently identified in the nitrogen fixing organism A. vinelandii (56). 5 The occurrence of nif-like genes in non-nitrogen fixing organisms and their counterparts in A. vinelandii led Dean and co-workers (56) to propose that these might play a role in formation or repair of iron-sulfur proteins and to designate them as isc (iron-sulfur cluster) genes. The sequential arrangement of the gene cluster, iscSUA-hscBA-fdx, is similar in the different organisms, and it appears likely that the genes are co-transcribed and encode proteins with coupled functions. The yeast genes identified in this study, SSQ1, JAC1, and NFS1, exhibit sequence similarity to hscA, hscB, and iscS, respec-tively. The similarity of Nfs1p to IscS suggests that it functions as a cysteine desulfurase. The fact that NFS1 is the only NifS/IscS-like gene in S. cerevisiae and the fact that it is essential for viability indicate that Nfs1p is likely to be the sole source for sulfide used for Fe/S cluster assembly. The roles of chaperone proteins, Ssq1p and Jac1p, are less obvious. Ssq1p exhibits ϳ40% identity to the hscA gene product, Hsc66, but a similar homology has also been noted between Hsc66 and the other mitochondrial hsp70 of yeast, Ssc1p (57). Jac1p, on the other hand, clearly seems to correspond to the yeast homologue of the hscB gene product, Hsc20. Several J-domain containing proteins are present in yeast, including mitochondrial Mdj1p (58) and Mdj2p (59), but Jac1p is the only one similar to Hsc20 in size and predicted domain structure. In vitro studies have shown that Hsc20 regulates the ATPase activity of the Hsc66 chaperone (60), and Jac1p may function in a similar manner as a co-chaperone to regulate the activity of Ssq1p. The findings that mutations in Ssq1p and Jac1p result in impaired respiratory rates and diminished activities of the mitochondrial Fe/S enzymes, aconitase and succinate dehydrogenase, suggest that these proteins may function in the assembly or repair of Fe/S clusters. Furthermore, mutations in SSQ1 have recently been shown to alter mitochondrial iron homeostasis (35).
In addition to iscS, hscA, and hscB, other members of the bacterial iscSUA-hscBA-fdx gene cluster appear to have counterparts in eukaryotes. Inspection of the S. cerevisiae genome reveals the presence of two homologues of iscU (YOR226C and YPL135W) and single homologues of iscA (YLL027W) and fdx (FDX1). Each of the predicted yeast proteins contains an Nterminal extension, not present in the corresponding bacterial protein, that exhibits features found in mitochondrial targeting sequences. Thus, eukaryotic mitochondria appear to have evolved with the same machinery for assembly of Fe/S clusters that is encoded by the iscSUA-hscBA-fdx gene cluster of bacteria. The majority of Fe/S proteins in eukaryotes are, in fact, located in mitochondria, and include proteins of the respiratory chain (NADH dehydrogenase, succinate dehydrogenase, Rieske protein) and citric acid cycle (aconitase). The findings presented here on the seo mutants are consistent with a model in which impairment of the mitochondrial Fe/S cluster assembly machinery leads to oxygen resistance in cells lacking copper/ zinc SOD1.
How might the status of mitochondrial iron-sulfur proteins affect oxygen resistance in sod1⌬ mutants? Three possibilities can be considered that are in agreement with available results. In one, decreased formation of Fe/S proteins of the respiratory chain is predicted to diminish O 2 . production. In S. cerevisiae, succinate dehydrogenase (complex II) and the Rieske protein of complex III comprise the respiratory chain Fe/S centers. 6 During respiration, succinate dehydrogenase and complex III activities affect steady state levels of ubisemiquinone, the primary site of reactive oxygen formation (61,62). Deficiencies in succinate dehydrogenase and/or the Rieske protein, by decreasing electron transfer rates, may thus decrease O 2 . formation.
Consistent with a decrease in electron transfer, all of our seo mutants exhibit diminished respiration rates. There is evidence that a fraction of SOD1 localizes to the intermembrane space of the mitochondria (12), and this enzyme may therefore help protect against O 2 . formed by the respiratory chain as previously suggested by Valentine and co-workers (21). A second possibility is that impairment of Fe/S cluster assembly in sod1⌬ mutants could reduce iron-related toxicity resulting from the oxidation of labile Fe/S centers. Keyer and Imlay (9) 5 D. R. Dean, personal communications. 6 Unlike mammalian systems, the NADH dehydrogenase of S. cerevisiae lacks an Fe/S center (65). have shown that cell damage in SOD-deficient E. coli results from oxidation of labile [4Fe-4S] clusters of dehydratase enzymes and subsequent release of free iron, and this mobilization of iron causes oxidative damage through increased hydroxyl radical formation (8). Yeast mitochondria harbor at least three enzymes predicted to contain labile [4Fe-4S] clusters, including two aconitase isoforms (Ref. 63 and accession number PO7263) and homoaconitase (64). Elevated levels of O 2 . in sod1⌬ mutants may oxidize these clusters, liberating iron that is then free to participate in Fenton chemistry either in the mitochondria or at other sites in the cell. Consistent with a role for SOD1 in protecting against the oxidation of mitochondrial [4Fe-4S] clusters, we observed a decrease in aconitase activity in sod1⌬ mutants. As a third possibility, defects in mitochondrial Fe/S cluster assembly could trigger a signaling pathway leading to changes in distribution of cytosolic and mitochondrial iron. Recent work by Dancis and colleagues (35) has shown that mutations in Ssq1 lead to hyperaccumulation of iron in mitochondria and an induction of high affinity iron uptake at the cell surface. This change in iron flux may help repair the labile Fe/S clusters damaged by oxidation in sod1⌬ mutants. Whether the mechanism involves direct leakage of respiratory superoxide or changes in iron homeostasis, these studies have established an important link between the status of Fe/S clusters in the mitochondria and Cu/Zn SOD1.